An Empirical Comparison of Regression Analysis Strategies with Discrete Ordinal Variables

Size: px
Start display at page:

Download "An Empirical Comparison of Regression Analysis Strategies with Discrete Ordinal Variables"

Transcription

1 Kromrey & Rendina-Gobioff An Empirical Comparison of Regression Analysis Strategies with Discrete Ordinal Variables Jeffrey D. Kromrey Gianna Rendina-Gobioff University of South Florida The Type I error control and statistical power of three tests of regression models incorporating discrete ordinal variables were compared in a Monte Carlo study. Samples were generated from populations measured on discrete ordinal variables representing 5-point and 7-point response scales. Each sample was analyzed using ordinary least squares regression, ordinal multiple regression and cumulative logit models. Factors examined in the Monte Carlo study were the population effect size, number of regressor variables, level of regressor intercorrelation, population distribution shape and sample size. Results suggest that the logistic regression approach evidenced poor Type I error control with small samples or with large numbers of regressors. In contrast, both the ordinary least squares approach and the ordinal multiple regression approach evidenced good Type I error control across the majority of conditions examined. Further, the power differences between these approaches were negligible. A common response format for the measurement of many variables in the social sciences is a forced-choice, ordinal scale. For example, data may be collected using a Likert scale in which respondents indicate the extent of their agreement or disagreement with a stimulus. Similarly, ordinal rating scales may be used to obtain ratings by participants regarding their perceptions of the frequency or intensity of a target phenomenon. The level of measurement of this type of variable and the appropriate statistical techniques for the analysis of such data have been the subject of debate for many years. Stevens (1951) defined four levels of measurement (nominal, ordinal, interval and ratio) and the appropriateness of their use within statistical analyses. The subsequently published literature presents conflicting views about the use of parametric statistics with such discrete ordinal variables. Critics of Stevens have proposed that ordinal data can be treated as interval data when parametric statistics are employed (e.g., Borgatta & Bohrnstedt, 1980; Gaito, 1980). In contrast, some researchers have proposed alternative analyses that are purported to more appropriately represent the discrete ordinal nature of such data (Agresti & Finlay, 1997; Cliff, 1994, 1996; Long, 1999), while others have recommended a rescaling of such data to provide a closer approximation to interval-level measurement (Harwell & Gatti, 2001). The long-standing debate about how to treat ordinal variables in statistical analyses is fueled by the need within the social sciences to accurately answer research questions in which measures are obtained using a discrete ordinal response scale. The purpose of this study was to investigate available analysis options for ordinal level data in the context of multiple regression analysis, and to empirically compare the performance of these analysis options. Multiple regression was the focus of this study because it has proven to be a useful, general purpose tool for data analysis encompassing the analysis of variance and analysis of covariance as special cases. As stated above, the literature reflects disagreement about appropriate analyses of discrete ordinal data. Therefore, one purpose of our work is to raise researchers awareness of multiple regression analysis options with discrete ordinal variables. Further, the literature base is lacking empirical studies that investigate factors of the research context (such as effect size, number of regressors, sample size, and level of measurement) that may affect the performance of the analysis options. Therefore, a second purpose was to compare the performance of three analysis options with various manipulations of factors under controlled conditions. Analysis Options with Discrete Ordinal Variables Through the years, ordinal data have been analyzed using a variety of models, including ordinary least squares regression, variants of logistic regression analysis, and techniques developed specifically for ordinal-level analyses. Each of these methods will be briefly described. Ordinary Least Squares. Ordinary least squares regression (OLS) analysis essentially ignores the discreteness and the ordinal level of measurement of variables, treating all values as if they represented a continuous interval-level measure. The well-known general linear model in matrix form is given by y = Xb+ ε, 30 Multiple Linear Regression Viewpoints, 2002, Vol. 28(2)

2 Discrete Ordinal Variables where y is an n x 1 vector of values for the dependent variable, X is an n x (k + 1) matrix of observations on the k regressor variables augmented with a unit vector to provide an intercept, b is a (k + 1) x1 vector of regression coefficients, and ε is an n x 1 vector of residuals. The OLS regression coefficients are obtained as: ˆ T = ( ) 1 T b X X X y. Hypothesis tests associated with OLS regression are typically of the form H 0 : β i = 0, where βi is the parameter associated with one of the elements of the vector b. This test is conducted by obtaining the ratio of the parameter estimate to its standard error and comparing that ratio to a t distribution with df = n k 1. A simultaneous test that all of the β i parameters are equal to zero (equivalent to the test that the population squared multiple correlation coefficient is equal to zero) is obtained by calculating the ratio 2 R F = k 2 ( 1 R ) ( n k 1 ) and comparing the result to an F distribution with df = k, n k 1. Ordinal Multiple Regression. Cliff (1994) presented an ordinal multiple regression strategy (OMR) and Long (1999) provided a method for estimating confidence intervals around the parameters of this model. The OMR model was developed to predict the ordinal information (that which Cliff refers to as dominance ) in a criterion variable, based upon the observed ordinal information among the predictor variables. That is, OMR predicts the dominance scores on the criterion variable (d ihy ) by weighting the observed dominance scores on the regressor variables (d ihx ). The dominance scores are given as dihx = sign( xi xh ), taking the value of 1 if the i th observation has a higher score on X than does the h th observation, a value of 1 if the score on X is lower, and a value of zero if the two observations are tied on X. The weights for the OMR model are obtained as 1 w = T x t y where w is a k X 1 vector of regression weights, T x is a k X k matrix of Kendall tau-a correlation coefficients among the regressors, and t y is a k X 1 vector of tau-a correlations between the regressors and the criterion variable. Confidence intervals and hypothesis tests about the OMR weights (w) are conducted using the standard normal (z) distribution. That is, the confidence interval for the j th weight, described by Long (1999) is w j ±Z (1-α/2) ˆ σ wj and a test of the null hypothesis that the population weight is zero is conducted by dividing the sample weight by its standard error and comparing the ratio to a critical value of Z. Logistic Regression. A cumulative logit model for ordinal criterion variables (LR) was described by Agresti and Finlay (1997). The LR analysis models the probability of a value on the dependent variable being in the j th category or lower, that is, modeling P( y j) based on the values of predictor variables. In logit form, the model is P( y j) log = α j + β X P( y > j) where α j is the intercept for the j th category of the ordinal response, and β is a k X 1 vector of regression weights for the predictors. Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 31

3 Kromrey & Rendina-Gobioff Parameter estimates for the LR models are usually obtained using maximum likelihood methods. Inference in the cumulative logit model is conducted using Wald tests for the individual regression parameters (ratios of the estimates to their standard errors), and likelihood ratio tests to obtain a simultaneous test that all parameters are equal to zero. Method This research was a Monte Carlo study in which random samples were generated under known and controlled population conditions. In the Monte Carlo study, samples were generated from populations of discrete ordinal variables. In each sample the data were analyzed using the OLS, OMR, and cumulative logit regression approaches. The Monte Carlo study included six factors in the design. These factors were (a) the true population effect size of the individual regressors (populations were simulated with effect sizes corresponding to small, medium and large values of Cohen s effect size f 2, as well as a null condition (effect size of zero), (b) number of regressor variables (with k = 2, 5 and 10 regessors), (c) correlation between regressor variables (with r 12 =.10 and.30), (d) sample sizes (with n = 5*k, 10*k and 100*k), (e) level of measurement of variables (discrete, ordinal variables were investigated as 5-option response scales, and 7- option response scales), and (f) population distribution shape (conditions were simulated in which the discrete ordinal variables were uniform, unimodal symmetric, and unimodal skewed). Generation of Pseudo-populations. Because the population parameters corresponding to regression weights in the models being investigated for discrete ordinal data cannot be determined analytically, the simulation study was conducted by generating a pseudo-population for each condition, then drawing random samples with replacement from these pseudo-populations. The pseudo-populations consisted of 10,000 observations randomly generated from the corresponding population (level of discreteness of the measurements, distribution shape and degree of relationship between the regressors and the criterion variable). The data were generated by transforming uniform random variates obtained from the RANUNI function in SAS, using a modification of the technique described by Bradley and Fleisher (1994), and operationalized by Ferron, Yi, and Kromrey (1997). In this method, a population correlation matrix, R, based on discrete ordinal variables is constructed by an iterative process in which large simulated samples R. The observed correlation matrix obtained (n = 100,000) are generated from an approximation of R, ( ) from this large sample ( ˆR ) is compared elementwise to R, and the residuals ( R R ˆ ) are used to adjust the generating matrix R. This sequence of large sample generation, matrix estimation, and adjustment of R continues until the process converges. The resulting matrix, R, is used as a template to generate correlated discrete data for the Monte Carlo study. Each analysis model was calculated on all observations in these pseudo-populations providing values that served as the corresponding population parameters. The Monte Carlo simulation was then conducted by randomly sampling observations, with replacement, from these pseudo-populations. The research was conducted using SAS/IML version 8.1. Conditions for the study were run under Windows 98. Discrete random variables were generated using the RANUNI function of SAS. A different seed value was used in each execution of the program and the program code was verified by handchecking results from benchmark datasets. For each condition investigated, 5,000 samples were generated. The use of 5,000 samples provides adequate precision for the investigation of the sampling behavior of these statistics. For example, 5,000 samples provides a maximum 95% confidence interval width around an observed proportion that is ±.014 (Robey & Barcikowski, 1992). The relative performance of the analysis strategies was evaluated by a comparison of the Type I error control and statistical power of the tests of regression coefficients. Estimates of Type I error control and statistical power were obtained by conducting the hypothesis tests associated with each parameter of each analytical model. For the OLS and LR models, the simultaneous test that all regression parameters is equal to zero was also conducted. The proportion of rejections of hypothesis tests for conditions in which 32 Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

4 Discrete Ordinal Variables Estimated Type I Error Rate OMR OLS LR Analysis Method FIGURE 1: Distribution of Type I Error Rate Estimates for Tests of Individual Regression Coefficients the population parameter is zero provided an estimate of Type I error control. Similarly, the proportion of rejections of such tests for conditions in which the parameter is not zero provided an estimate of statistical power. Results Type I Error Control Box-and-whisker plots in Figure 1 depict the distribution of Type I error rate estimates for tests of individual regression weights across the conditions examined in the research. The figure clearly portrays a higher typical Type I error rate for the LR analysis method, relative to the OMR and OLS methods. The tails of the plot for the LR method range from.02 to.12, with half ranging from.06 to.08. The OMR method exhibits the most consistent Type I estimates, with the majority of estimates at.06. The OLS method is similar to the OMR with tails ranging from.04 to.07, with half ranging from.05 to.06. Estimates of Type I error rates for the tests of individual regression weights in samples from the unimodal symmetric population are presented in Table 1. Overall the 5-point scale and 7-point scale exhibited few differences across the three analysis methods for all conditions. With two regressors and small sample size, the Type I error rate for the LR analysis method was lower than the OMR and OLS analysis methods. For example, with a 5-point scale, k=2, r 12 =.30, and n=10, the estimated Type I error rate for OMR was.07, for OLS was.05, and for LR was.04. However, with more than 2 regressors and small to medium sample sizes, the Type I error rate was higher for LR when compared to OLS and OMR. For example, with a 5-point scale, k=10, r 12 =.10, and n=50, the estimated Type I error rate for LR (.09) was higher than OMR (.05) and OLS (.05). The OMR method had slightly higher Type I error rates with 2 regressors when compared to its performance with 5 and 10 regressors. The analysis methods provided similar results across models with five and ten regressors, with LR consistently having higher Type I error rates than OMR or OLS. The OMR and OLS methods for small, medium, and larger sample sizes exhibited similar Type I error rates, which ranged from.05 to.07. In contrast the LR analysis method for small and medium sample sizes exhibited a higher Type I error rate, Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 33

5 Kromrey & Rendina-Gobioff Table 1. Type I Error Rate Estimates for Tests of Individual Regression Weights with Unimodal Symmetric Populations 5-point Scale Analysis Method k r 12 N OMR OLS LR point Scale Analysis Method k r 12 N OMR OLS LR which ranged from.06 to.10. For the small sample conditions, the LR method provided slightly higher Type I error rates with 10 regressors than with 5 regressors (although it remained liberal with small samples in both conditions). For example, with a 7-point scale, k=10, r 12 =.30, and n=50, the estimated Type I error rate for LR was.10. In contrast, with k = 5, r 12 =.30, and n = 25, the Type I error rate estimate for LR dropped to.09. The LR method evidenced better Type I error control as the sample size increased. For example, with a 7-point scale, k=5, r 12 =.30, the LR method was.09 with the small sample,.08 with the medium sample, and.05 with the large sample. The Type I error rate estimates for tests of individual regression weights in selected conditions across distribution shapes are presented in Table 2. The three distribution shapes (unimodal symmetric, uniform, 34 Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

6 Discrete Ordinal Variables Table 2. Type I Error Rates Estimates for Tests of Individual Regression Weights Across Distribution Shapes. 5-point Scale Symmetric Uniform Skewed k r 12 N OMR OLS LR OMR OLS LR OMR OLS LR point Scale Symmetric Uniform Skewed k r 12 N OMR OLS LR OMR OLS LR OMR OLS LR and skewed) yielded similar Type I error rates for all three analysis methods. For example, with the 7- point scale, k=5, r 12 =.10, and n=25, the estimated Type I error rate ranged from.05 to.06 across the distribution shapes for the OMR and OLS analysis method. Similarly, for the LR analysis method the estimated Type I error rate ranged from.08 to.09 across the distribution shapes. Figure 2 contains box-and-whisker plots depicting the distribution of Type I error rate estimates for simultaneous tests of regression coefficients. The LR method portrays a wide range of estimates,.00 to.18, with half ranging from.00 to.09. The OLS method portrays a much smaller range of estimates,.04 to.07, with half ranging from.05 to.06. Thus the OLS method exhibits a much smaller type I error rate for simultaneous tests of regression coefficients across all research conditions. Table 3 presents the estimates of Type I error rates for the simultaneous test of all regression weights when samples were drawn from the unimodal symmetric populations. As evident in this table, the LR method evidenced substantially greater variability in Type I error control for this test than did the OLS method. Specifically, the LR method was very conservative in conditions with both k = 2 and k = 5 and was liberal with k = 10, with Type I error rates reaching as high as.14 (with the 7-point scale, k = 10, r 12 =.3, and n=50). The OLS method maintained Type I error rates near the nominal level across these conditions, with larger estimates in the larger sample size and 10k conditions. The Type I error rates for these tests across distribution shapes (Table 4) indicated a slight increase in Type I error rates for the OLS method when samples were drawn from the 7 point, 10 regressors, symmetric and skewed distributions, relative to the rates obtained from the uniform distribution. However, these higher rates remained very close to the nominal alpha level and did not exceed.09. Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 35

7 Kromrey & Rendina-Gobioff Estimated Type I Error Rate OLS LR Analysis Method FIGURE 2: Distribution of Type I Error Rate Estimates for Simultaneous Test of Regression Coefficients Statistical Power The distribution of estimates of statistical power for the tests of regression weights across the conditions examined in this research are presented as box-and-whisker plots in Figure 3. As is evident in this figure, most of the conditions examined yielded relatively low power and the power differences across analysis methods was not great. However, the LR analysis method provided slight power advantages relative to OMR and OLS analyses (a result that should be expected because of its tendency to be liberal in Type I error control). Power estimates for the tests of individual regression weights with samples drawn from the three distribution shapes are presented in Tables 5 and 6. Few differences were evident between the analyses of variables with 5 points and those with 7 points. With small effect sizes (f 2 =.02) the power estimates for all analysis methods ranged from.04 to.12. With large effect sizes (f 2 =.35), modest but greater power differences were evident across the analysis methods. With a small number of regressors (k = 2), the OMR method evidenced power advantages over the OLS and LR methods with small samples. With a 5- point scale, f 2 =.35, k = 2, r 12 =.10, and n = 10, for example, the statistical power of the OMR method was estimated to be.19, while the power estimates of the OLS and LR methods were.15 and.09, respectively (Table 5). These slight power advantages of the OMR method disappeared with larger samples and larger numbers of regressors (conditions in which the LR approach evidenced modest power advantages). The differences in statistical power between tests conducted on samples from unimodal symmetric, uniform, and skewed distributions are negligible. For example, with the 7-point response scale, f 2 =.35, k = 5, r 12 =.10, and n = 50, the power estimates for the OMR method ranged from.21 for the skewed population to.28 for the uniform population (Table 6). Similarly, the power estimates for OLS ranged only from.28 to.31, and those of the LR method ranged from.32 to.35. The distribution of power estimates for simultaneous tests of regression coefficients across all research conditions is presented with box-and-whisker plots in Figure 4. The OLS method evidenced a slight power advantage over the LR method. Power estimates for the simultaneous test of all regression 36 Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

8 Discrete Ordinal Variables Table 3. Estimates of Type I Error Rate for Simultaneous Tests of All Regression Weights with Unimodal Symmetric Populations 5-Point Scale 7-Point Scale Analysis Method Analysis Method k r 12 N OLS LR OLS LR weights with samples drawn from unimodal symmetric, uniform, and skewed populations are presented in Tables 7 and 8. With a small effect size (.02) and k = 2 or k = 5, the LR test provided rejection rates less than the nominal alpha level across most conditions. For the OLS test, similarly low power levels were observed except for large samples (and power estimates were still less than.50 for these conditions). As expected, with larger effect sizes the power estimates for both methods increased. The OLS method evidenced substantially greater power than the LR method for k = 2 and k = 5, while the LR method evidenced smaller power advantages for k = 10. A comparison of power estimates across distribution shapes suggests that population shape is not a major influence on the power of either the OLS or the LR method. Conclusions Surprisingly few and relatively small differences were evident among the OMR, OLS and LR methods in terms of their Type I error control and statistical power in tests of their respective weights. The OMR approach, while accurately representing the ordinal nature of the discrete response scales and discrete regressor variables, provided neither superior Type I error control nor superior statistical power. In contrast, the LR analysis evidenced Type I error control problems with small sample sizes or large numbers of regressors, conditions in which the Wald tests of the logistic regression weights became liberal. Although this analysis method presents modest power advantages relative to OMR and OLS, such power comes at a cost of relatively tenuous Type I error control. Finally, the surprisingly good performance of the OLS approach suggests that researchers who approach the analysis of discrete ordinal data (such as individual Likert items) with OLS tools should feel no guilt in such a tactic. The Type I error control in tests conducted of the OLS regression weights was as good as that obtained with tests of OMR weights and was superior to tests obtained in the LR context. Further, the statistical power evidenced with OLS was comparable or superior to that of OMR. Of course, testing hypotheses about weights obtained in these models are only a small part of the inferential machinery applied to discrete ordinal data. Additional research is needed to focus on the relative bias in sample estimates of weights for OMR, OLS, and LR models and on the accuracy of confidence bands constructed around the sample estimated weights. Furthermore, research conducted on models that present a mixture of discrete ordinal variables and continuous variables is needed in order to explore the relative performance of these models in such a context. Finally, the substantive nature of the Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 37

9 Kromrey & Rendina-Gobioff Table 4. Type I Error Rate Estimates for Simultaneous Tests of All Regression Weights Across Distribution Shapes. 5-point Scale Symmetric Uniform Skewed k r 12 N OLS LR OLS LR OLS LR point Scale Symmetric Uniform Skewed k r 12 N OLS LR OLS LR OLS LR inferences suggested by these models requires attention from the perspective of psychometrics. That is, the OMR model provides a vehicle for inferences about the ordinal position of observations on the criterion variable, while the LR model estimates the probability of an observation being in category j or lower on the criterion variable. In contrast, the OLS model provides a prediction of the criterion value as though it was a continuous variable (i.e., seeking to minimize the sum of squared errors of prediction). In addition (although not analyzed in this research study), methods for determining both the strength and direction of the response have been proposed (Jones & Sobel (2000); Brody & Dietz (1997). These types of inferences have obvious substantive differences, and the validity of such inferences extends beyond the estimation of Type I error rates and statistical power. In conclusion, this study has sought to bring increased awareness and clarity to three analysis options for multiple regression with discrete ordinal variables. The social sciences often focus on variables which are measured with ordinal scales (most commonly Likert scales). Unfortunately, the appropriate application of multiple regression with discrete ordinal data has been insufficiently addressed in the literature. There are many reasons for this paucity of treatment of the issue: general debates continue about the treatment of ordinal data with statistics, the breadth of analysis options are spread among numerous sources, and the consequences of analysis choice in terms of Type I error control and statistical power has not been thoroughly investigated. The current study presents three analysis options within the literature and clarifies their differences with the hopes of both increasing the appropriate analysis of discrete ordinal variables and stimulating additional methodological research in this area. 38 Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

10 Discrete Ordinal Variables Estimated Statistical Power OMR OLS LR Analysis Method FIGURE 3: Distribution of Statistical Power Estimates for Tests of Individual Regression Coefficients Estimated Statistical Power OLS LR Analysis Method FIGURE 4: Distribution of Statistical Power Estimates for Simultaneous Test of Regression Coefficients Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 39

11 Kromrey & Rendina-Gobioff Table 5. Estimates of Statistical Power for Tests of Individual Regression Weights Across Distribution Shapes. 5 Point Likert Symmetric Uniform Skewed Effect k r 12 N OMR OLS LR OMR OLS LR OMR OLS LR References Agresti, A. & Finlay, B. (1997). Statistical methods for the social sciences (3rd Ed.). Upper Saddle River, NJ: Prentice-Hall. Borgatta, E. F. & Bohrnstedt, G. W. (1980). Level of measurement: Once over again. Sociological Methods & Research, 9(2), Bradley, D. R. & Fleisher, C. L. (1994). Generating multivariate data from nonnormal distributions: Mihal and Barrett revisited. Behavior Research Methods, Instruments, and Computers, 26, Brody, C. J. & Dietz, J. (1997). On the dimensionality of two-question format likert attitude scales. Social Science Research, 26, Cliff, N. (1994). Predicting ordinal relations. British Journal of Mathematical and Statistical Psychology, 47, Cliff, N. (1996). Ordinal Methods for Behavioral Data Analysis. Mahwah, NJ: Lawrence Erlbaum Associates. Ferron, J., Yi, Q. & Kromrey, J.D. (1997). NNCORR: A SAS/IML program for generating nonnormal correlated data. Applied Psychological Measurement, 21, Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

12 Discrete Ordinal Variables Table 6. Estimates of Statistical Power for Tests of Individual Regression Weights Across Distribution Shapes. 7 point Likert Symmetric Uniform Skewed Effect k r 12 N OMR OLS LR OMR OLS LR OMR OLS LR Gaito, J. (1980). Measurement scales and statistics: Resurgence of an old misconception. Psychological Bulletin, 87(3), Harwell, M. R. & Gatti, G. G. (2001). Rescaling ordinal data to interval data in educational research. Review of Educational Research, 71, Jones, B. S. & Sobel, M. E. (2000). Modeling direction and intensity in semantically balanced ordinal scales: An assessment of congressional incumbent approval. American Journal of Political Science, 44(1), Long, J. D. (1999). A confidence interval for ordinal multiple regression weights. Psychological Methods, 4(3), Stevens, S. S. (1951). Mathematics, measurement and psychophysics. In S. S. Stevens (Ed.), Handbook of experimental psychology New York: Wiley. Robey, R. R. & Barcikowski, R. S. (1992). Type I error and the number of iterations in Monte Carlo studies of robustness. British Journal of Mathematical and Statistical Psychology, 45, Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 41

13 Kromrey & Rendina-Gobioff Table 7. Estimates of Statistical Power for Simultaneous Tests of All Regression Weights Across Distribution Shapes. 5 Point Likert Symmetric Uniform Skewed Effect k r 12 N OLS LR OLS LR OLS LR Multiple Linear Regression Viewpoints, 2003, Vol. 29(1)

14 Discrete Ordinal Variables Table 8. Estimates of Statistical Power for Simultaneous Tests of All Regression Weights Across Distribution Shapes. 7 Point Likert Symmetric Uniform Skewed Effect k r 12 N OLS LR OLS LR OLS LR Send correspondence to: Jeffrey D. Kromrey Department of Educational Measurement and Research University of South Florida 4202 E. Fowler Ave., EDU162, Tampa, FL Kromrey@tempest.coedu.usf.edu Multiple Linear Regression Viewpoints, 2003, Vol. 29(1) 43

Legendre et al Appendices and Supplements, p. 1

Legendre et al Appendices and Supplements, p. 1 Legendre et al. 2010 Appendices and Supplements, p. 1 Appendices and Supplement to: Legendre, P., M. De Cáceres, and D. Borcard. 2010. Community surveys through space and time: testing the space-time interaction

More information

Running head: DATA ANALYSIS AND INTERPRETATION 1

Running head: DATA ANALYSIS AND INTERPRETATION 1 Running head: DATA ANALYSIS AND INTERPRETATION 1 Data Analysis and Interpretation Final Project Vernon Tilly Jr. University of Central Oklahoma DATA ANALYSIS AND INTERPRETATION 2 Owners of the various

More information

Chapter 12 Practice Test

Chapter 12 Practice Test Chapter 12 Practice Test 1. Which of the following is not one of the conditions that must be satisfied in order to perform inference about the slope of a least-squares regression line? (a) For each value

More information

Name May 3, 2007 Math Probability and Statistics

Name May 3, 2007 Math Probability and Statistics Name May 3, 2007 Math 341 - Probability and Statistics Long Exam IV Instructions: Please include all relevant work to get full credit. Encircle your final answers. 1. An article in Professional Geographer

More information

Section I: Multiple Choice Select the best answer for each problem.

Section I: Multiple Choice Select the best answer for each problem. Inference for Linear Regression Review Section I: Multiple Choice Select the best answer for each problem. 1. Which of the following is NOT one of the conditions that must be satisfied in order to perform

More information

Introduction to Analysis of Variance (ANOVA) The Structural Model, The Summary Table, and the One- Way ANOVA

Introduction to Analysis of Variance (ANOVA) The Structural Model, The Summary Table, and the One- Way ANOVA Introduction to Analysis of Variance (ANOVA) The Structural Model, The Summary Table, and the One- Way ANOVA Limitations of the t-test Although the t-test is commonly used, it has limitations Can only

More information

Robust specification testing in regression: the FRESET test and autocorrelated disturbances

Robust specification testing in regression: the FRESET test and autocorrelated disturbances Robust specification testing in regression: the FRESET test and autocorrelated disturbances Linda F. DeBenedictis and David E. A. Giles * Policy and Research Division, Ministry of Human Resources, 614

More information

Using Actual Betting Percentages to Analyze Sportsbook Behavior: The Canadian and Arena Football Leagues

Using Actual Betting Percentages to Analyze Sportsbook Behavior: The Canadian and Arena Football Leagues Syracuse University SURFACE College Research Center David B. Falk College of Sport and Human Dynamics October 2010 Using Actual Betting s to Analyze Sportsbook Behavior: The Canadian and Arena Football

More information

Evaluation of Regression Approaches for Predicting Yellow Perch (Perca flavescens) Recreational Harvest in Ohio Waters of Lake Erie

Evaluation of Regression Approaches for Predicting Yellow Perch (Perca flavescens) Recreational Harvest in Ohio Waters of Lake Erie Evaluation of Regression Approaches for Predicting Yellow Perch (Perca flavescens) Recreational Harvest in Ohio Waters of Lake Erie QFC Technical Report T2010-01 Prepared for: Ohio Department of Natural

More information

Data Set 7: Bioerosion by Parrotfish Background volume of bites The question:

Data Set 7: Bioerosion by Parrotfish Background volume of bites The question: Data Set 7: Bioerosion by Parrotfish Background Bioerosion of coral reefs results from animals taking bites out of the calcium-carbonate skeleton of the reef. Parrotfishes are major bioerosion agents,

More information

Announcements. Lecture 19: Inference for SLR & Transformations. Online quiz 7 - commonly missed questions

Announcements. Lecture 19: Inference for SLR & Transformations. Online quiz 7 - commonly missed questions Announcements Announcements Lecture 19: Inference for SLR & Statistics 101 Mine Çetinkaya-Rundel April 3, 2012 HW 7 due Thursday. Correlation guessing game - ends on April 12 at noon. Winner will be announced

More information

A Case Study of Leadership in Women s Intercollegiate Softball. By: DIANE L. GILL and JEAN L. PERRY

A Case Study of Leadership in Women s Intercollegiate Softball. By: DIANE L. GILL and JEAN L. PERRY A Case Study of Leadership in Women s Intercollegiate Softball By: DIANE L. GILL and JEAN L. PERRY Gill, D.L. & Perry, J.L. (1979). A case study of leadership in women's intercollegiate softball. International

More information

What Causes the Favorite-Longshot Bias? Further Evidence from Tennis

What Causes the Favorite-Longshot Bias? Further Evidence from Tennis MPRA Munich Personal RePEc Archive What Causes the Favorite-Longshot Bias? Further Evidence from Tennis Jiri Lahvicka 30. June 2013 Online at http://mpra.ub.uni-muenchen.de/47905/ MPRA Paper No. 47905,

More information

Stat 139 Homework 3 Solutions, Spring 2015

Stat 139 Homework 3 Solutions, Spring 2015 Stat 39 Homework 3 Solutions, Spring 05 Problem. Let i Nµ, σ ) for i,..., n, and j Nµ, σ ) for j,..., n. Also, assume that all observations are independent from each other. In Unit 4, we learned that the

More information

Determination of the Design Load for Structural Safety Assessment against Gas Explosion in Offshore Topside

Determination of the Design Load for Structural Safety Assessment against Gas Explosion in Offshore Topside Determination of the Design Load for Structural Safety Assessment against Gas Explosion in Offshore Topside Migyeong Kim a, Gyusung Kim a, *, Jongjin Jung a and Wooseung Sim a a Advanced Technology Institute,

More information

Modeling Pedestrian Volumes on College Campuses

Modeling Pedestrian Volumes on College Campuses TRANSPORTATION RESEARCH RECORD 1405 43 Modeling Pedestrian Volumes on College Campuses LAURAL. COVE AND J. EDWIN CLARK A study was undertaken to develop a reliable method for obtaining reasonable estimates

More information

Journal of Quantitative Analysis in Sports Manuscript 1039

Journal of Quantitative Analysis in Sports Manuscript 1039 An Article Submitted to Journal of Quantitative Analysis in Sports Manuscript 1039 A Simple and Flexible Rating Method for Predicting Success in the NCAA Basketball Tournament Brady T. West University

More information

Addendum to SEDAR16-DW-22

Addendum to SEDAR16-DW-22 Addendum to SEDAR16-DW-22 Introduction Six king mackerel indices of abundance, two for each region Gulf of Mexico, South Atlantic, and Mixing Zone, were constructed for the SEDAR16 data workshop using

More information

Two Machine Learning Approaches to Understand the NBA Data

Two Machine Learning Approaches to Understand the NBA Data Two Machine Learning Approaches to Understand the NBA Data Panagiotis Lolas December 14, 2017 1 Introduction In this project, I consider applications of machine learning in the analysis of nba data. To

More information

Distancei = BrandAi + 2 BrandBi + 3 BrandCi + i

Distancei = BrandAi + 2 BrandBi + 3 BrandCi + i . Suppose that the United States Golf Associate (USGA) wants to compare the mean distances traveled by four brands of golf balls when struck by a driver. A completely randomized design is employed with

More information

Analysis of Variance. Copyright 2014 Pearson Education, Inc.

Analysis of Variance. Copyright 2014 Pearson Education, Inc. Analysis of Variance 12-1 Learning Outcomes Outcome 1. Understand the basic logic of analysis of variance. Outcome 2. Perform a hypothesis test for a single-factor design using analysis of variance manually

More information

Analysis of Shear Lag in Steel Angle Connectors

Analysis of Shear Lag in Steel Angle Connectors University of New Hampshire University of New Hampshire Scholars' Repository Honors Theses and Capstones Student Scholarship Spring 2013 Analysis of Shear Lag in Steel Angle Connectors Benjamin Sawyer

More information

Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework

Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework Calhoun: The NPS Institutional Archive DSpace Repository Faculty and Researchers Faculty and Researchers Collection 2010 Revisiting the Hot Hand Theory with Free Throw Data in a Multivariate Framework

More information

Novel empirical correlations for estimation of bubble point pressure, saturated viscosity and gas solubility of crude oils

Novel empirical correlations for estimation of bubble point pressure, saturated viscosity and gas solubility of crude oils 86 Pet.Sci.(29)6:86-9 DOI 1.17/s12182-9-16-x Novel empirical correlations for estimation of bubble point pressure, saturated viscosity and gas solubility of crude oils Ehsan Khamehchi 1, Fariborz Rashidi

More information

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD Outline Definition. Deriving the Estimates. Properties of the Estimates. Units of Measurement and Functional Form. Expected

More information

Driver Behavior at Highway-Rail Grade Crossings With Passive Traffic Controls

Driver Behavior at Highway-Rail Grade Crossings With Passive Traffic Controls 2014 Global Level Crossing Symposium August 2014, Urbana, IL, USA Driver Behavior at Highway-Rail Grade Crossings With Passive Traffic Controls - A Driving Simulator Study Presenter: Dr. Asad J. Khattak

More information

Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League.

Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League. MPRA Munich Personal RePEc Archive Emergence of a professional sports league and human capital formation for sports: The Japanese Professional Football League. Eiji Yamamura 26. February 2013 Online at

More information

Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix

Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix Measuring Returns to Scale in Nineteenth-Century French Industry Technical Appendix Ulrich Doraszelski Hoover Institution, Stanford University March 2004 Formal Derivations Gross-output vs value-added

More information

Navigate to the golf data folder and make it your working directory. Load the data by typing

Navigate to the golf data folder and make it your working directory. Load the data by typing Golf Analysis 1.1 Introduction In a round, golfers have a number of choices to make. For a particular shot, is it better to use the longest club available to try to reach the green, or would it be better

More information

Stats 2002: Probabilities for Wins and Losses of Online Gambling

Stats 2002: Probabilities for Wins and Losses of Online Gambling Abstract: Jennifer Mateja Andrea Scisinger Lindsay Lacher Stats 2002: Probabilities for Wins and Losses of Online Gambling The objective of this experiment is to determine whether online gambling is a

More information

PSY201: Chapter 5: The Normal Curve and Standard Scores

PSY201: Chapter 5: The Normal Curve and Standard Scores PSY201: Chapter 5: The Normal Curve and Standard Scores Introduction: Normal curve + a very important distribution in behavior sciences + three principal reasons why... - 1. many of the variables measured

More information

Extreme Shooters in the NBA

Extreme Shooters in the NBA Extreme Shooters in the NBA Manav Kant and Lisa R. Goldberg 1. Introduction December 24, 2018 It is widely perceived that star shooters in basketball have hot and cold streaks. This perception was first

More information

CHAPTER ANALYSIS AND INTERPRETATION Average total number of collisions for a try to be scored

CHAPTER ANALYSIS AND INTERPRETATION Average total number of collisions for a try to be scored CHAPTER 8 8.1 ANALYSIS AND INTERPRETATION As mentioned in the previous chapter, four key components have been identified as indicators of the level of significance of dominant collisions when evaluating

More information

POWER Quantifying Correction Curve Uncertainty Through Empirical Methods

POWER Quantifying Correction Curve Uncertainty Through Empirical Methods Proceedings of the ASME 2014 Power Conference POWER2014 July 28-31, 2014, Baltimore, Maryland, USA POWER2014-32187 Quantifying Correction Curve Uncertainty Through Empirical Methods ABSTRACT Christopher

More information

An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings

An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings An Application of Signal Detection Theory for Understanding Driver Behavior at Highway-Rail Grade Crossings Michelle Yeh and Jordan Multer United States Department of Transportation Volpe National Transportation

More information

The Reliability of Intrinsic Batted Ball Statistics Appendix

The Reliability of Intrinsic Batted Ball Statistics Appendix The Reliability of ntrinsic Batted Ball Statistics Appendix Glenn Healey, EECS Department University of California, rvine, CA 92617 Given information about batted balls for a set of players, we review

More information

Real-Time Electricity Pricing

Real-Time Electricity Pricing Real-Time Electricity Pricing Xi Chen, Jonathan Hosking and Soumyadip Ghosh IBM Watson Research Center / Northwestern University Yorktown Heights, NY, USA X. Chen, J. Hosking & S. Ghosh (IBM) Real-Time

More information

United States Commercial Vertical Line Vessel Standardized Catch Rates of Red Grouper in the US South Atlantic,

United States Commercial Vertical Line Vessel Standardized Catch Rates of Red Grouper in the US South Atlantic, SEDAR19-DW-14 United States Commercial Vertical Line Vessel Standardized Catch Rates of Red Grouper in the US South Atlantic, 1993-2008 Kevin McCarthy and Neil Baertlein National Marine Fisheries Service,

More information

Keywords: multiple linear regression; pedestrian crossing delay; right-turn car flow; the number of pedestrians;

Keywords: multiple linear regression; pedestrian crossing delay; right-turn car flow; the number of pedestrians; Available online at www.sciencedirect.com ScienceDirect Procedia - Social and Behavioral Scien ce s 96 ( 2013 ) 1997 2003 13th COTA International Conference of Transportation Professionals (CICTP 2013)

More information

Clutch Hitters Revisited Pete Palmer and Dick Cramer National SABR Convention June 30, 2008

Clutch Hitters Revisited Pete Palmer and Dick Cramer National SABR Convention June 30, 2008 Clutch Hitters Revisited Pete Palmer and Dick Cramer National SABR Convention June 30, 2008 Do clutch hitters exist? More precisely, are there any batters whose performance in critical game situations

More information

NCSS Statistical Software

NCSS Statistical Software Chapter 256 Introduction This procedure computes summary statistics and common non-parametric, single-sample runs tests for a series of n numeric, binary, or categorical data values. For numeric data,

More information

Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes

Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes Deciding When to Quit: Reference-Dependence over Slot Machine Outcomes By JAIMIE W. LIEN * AND JIE ZHENG * Lien: Department of Economics, School of Economics and Management, Tsinghua University, Beijing,

More information

A few things to remember about ANOVA

A few things to remember about ANOVA A few things to remember about ANOVA 1) The F-test that is performed is always 1-tailed. This is because your alternative hypothesis is always that the between group variation is greater than the within

More information

Journal of Human Sport and Exercise E-ISSN: Universidad de Alicante España

Journal of Human Sport and Exercise E-ISSN: Universidad de Alicante España Journal of Human Sport and Exercise E-ISSN: 1988-5202 jhse@ua.es Universidad de Alicante España SOÓS, ISTVÁN; FLORES MARTÍNEZ, JOSÉ CARLOS; SZABO, ATTILA Before the Rio Games: A retrospective evaluation

More information

Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias

Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias Behavior under Social Pressure: Empty Italian Stadiums and Referee Bias Per Pettersson-Lidbom a and Mikael Priks bc* April 11, 2010 Abstract Due to tightened safety regulation, some Italian soccer teams

More information

SEDAR52-WP November 2017

SEDAR52-WP November 2017 Using a Censored Regression Modeling Approach to Standardize Red Snapper Catch per Unit Effort Using Recreational Fishery Data Affected by a Bag Limit Skyler Sagarese and Adyan Rios SEDAR52-WP-13 15 November

More information

Competitive Performance of Elite Olympic-Distance Triathletes: Reliability and Smallest Worthwhile Enhancement

Competitive Performance of Elite Olympic-Distance Triathletes: Reliability and Smallest Worthwhile Enhancement SPORTSCIENCE sportsci.org Original Research / Performance Competitive Performance of Elite Olympic-Distance Triathletes: Reliability and Smallest Worthwhile Enhancement Carl D Paton, Will G Hopkins Sportscience

More information

Has the NFL s Rooney Rule Efforts Leveled the Field for African American Head Coach Candidates?

Has the NFL s Rooney Rule Efforts Leveled the Field for African American Head Coach Candidates? University of Pennsylvania ScholarlyCommons PSC Working Paper Series 7-15-2010 Has the NFL s Rooney Rule Efforts Leveled the Field for African American Head Coach Candidates? Janice F. Madden University

More information

A Cost Effective and Efficient Way to Assess Trail Conditions: A New Sampling Approach

A Cost Effective and Efficient Way to Assess Trail Conditions: A New Sampling Approach A Cost Effective and Efficient Way to Assess Trail Conditions: A New Sampling Approach Rachel A. Knapp, Graduate Assistant, University of New Hampshire Department of Natural Resources and the Environment,

More information

The final set in a tennis match: four years at Wimbledon 1

The final set in a tennis match: four years at Wimbledon 1 Published as: Journal of Applied Statistics, Vol. 26, No. 4, 1999, 461-468. The final set in a tennis match: four years at Wimbledon 1 Jan R. Magnus, CentER, Tilburg University, the Netherlands and Franc

More information

How Many Iterations Are Enough?

How Many Iterations Are Enough? T ECOLOTE R ESEARCH, I NC. Bridging Engineering and Economics Since 1973 How Many Are Enough? Alfred Smith 2008 Joint SCEA/ISPA Annual Conference and Training Workshop Pacific Palms Conference Resort,

More information

SPATIAL STATISTICS A SPATIAL ANALYSIS AND COMPARISON OF NBA PLAYERS. Introduction

SPATIAL STATISTICS A SPATIAL ANALYSIS AND COMPARISON OF NBA PLAYERS. Introduction A SPATIAL ANALYSIS AND COMPARISON OF NBA PLAYERS KELLIN RUMSEY Introduction The 2016 National Basketball Association championship featured two of the leagues biggest names. The Golden State Warriors Stephen

More information

Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments

Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments Nathaniel Beck Jonathan N. Katz Draft of April 16, 2006 We gratefully acknowledge the financial support of the National

More information

DOCUMENT RESUME. A Comparison of Type I Error Rates of Alpha-Max with Established Multiple Comparison Procedures. PUB DATE NOTE

DOCUMENT RESUME. A Comparison of Type I Error Rates of Alpha-Max with Established Multiple Comparison Procedures. PUB DATE NOTE DOCUMENT RESUME ED 415 284 TM 028 030 AUTHOR Barnette, J. Jackson; McLean, James E. TITLE A Comparison of Type I Error Rates of Alpha-Max with Established Multiple Comparison Procedures. PUB DATE 1997-11-13

More information

An Empirical Analysis of the Impact of Renewable Portfolio Standards and Feed-in-Tariffs on International Markets.

An Empirical Analysis of the Impact of Renewable Portfolio Standards and Feed-in-Tariffs on International Markets. An Empirical Analysis of the Impact of Renewable Portfolio Standards and Feed-in-Tariffs on International Markets. Presentation by Greg Upton Gregory Upton Jr. Louisiana State University Sanya Carley Indiana

More information

Bhagwant N. Persaud* Richard A. Retting Craig Lyon* Anne T. McCartt. May *Consultant to the Insurance Institute for Highway Safety

Bhagwant N. Persaud* Richard A. Retting Craig Lyon* Anne T. McCartt. May *Consultant to the Insurance Institute for Highway Safety Review of The Impact of Red Light Cameras (Photo-Red Enforcement) on Crashes in Virginia by Nicholas J. Garber, John S. Miller, R. Elizabeth Abel, Saeed Eslambolchi, and Santhosh K. Korukonda Bhagwant

More information

To Illuminate or Not to Illuminate: Roadway Lighting as It Affects Traffic Safety at Intersections

To Illuminate or Not to Illuminate: Roadway Lighting as It Affects Traffic Safety at Intersections To Illuminate or Not to Illuminate: Roadway Lighting as It Affects Traffic Safety at Intersections Mark Rea Lighting Research Center Rensselaer Polytechnic Institute Eric Donnell Dept. of Civil and Environmental

More information

E STIMATING KILN SCHEDULES FOR TROPICAL AND TEMPERATE HARDWOODS USING SPECIFIC GRAVITY

E STIMATING KILN SCHEDULES FOR TROPICAL AND TEMPERATE HARDWOODS USING SPECIFIC GRAVITY P ROCESSES E STIMATING KILN SCHEDULES FOR TROPICAL AND TEMPERATE HARDWOODS USING SPECIFIC GRAVITY W ILLIAM T. SIMPSON S TEVE P. VERRILL A BSTRACT Dry-kiln schedules have been developed for many hardwood

More information

Staking plans in sports betting under unknown true probabilities of the event

Staking plans in sports betting under unknown true probabilities of the event Staking plans in sports betting under unknown true probabilities of the event Andrés Barge-Gil 1 1 Department of Economic Analysis, Universidad Complutense de Madrid, Spain June 15, 2018 Abstract Kelly

More information

Modeling the NCAA Tournament Through Bayesian Logistic Regression

Modeling the NCAA Tournament Through Bayesian Logistic Regression Duquesne University Duquesne Scholarship Collection Electronic Theses and Dissertations 0 Modeling the NCAA Tournament Through Bayesian Logistic Regression Bryan Nelson Follow this and additional works

More information

Math 121 Test Questions Spring 2010 Chapters 13 and 14

Math 121 Test Questions Spring 2010 Chapters 13 and 14 Math 121 Test Questions Spring 2010 Chapters 13 and 14 1. (10 pts) The first-semester enrollment at HSC was 1120 students. If all entering freshmen classes were the same size and there were no attrition,

More information

Modelling residential prices with cointegration techniques and automatic selection algorithms

Modelling residential prices with cointegration techniques and automatic selection algorithms Modelling residential prices with cointegration techniques and automatic selection algorithms Ramiro J. Rodríguez A presentation for ERES 2014 - PhD Sessions Bucharest, Rumania The opinions and analyses

More information

Economic Value of Celebrity Endorsements:

Economic Value of Celebrity Endorsements: Economic Value of Celebrity Endorsements: Tiger Woods Impact on Sales of Nike Golf Balls September 27, 2012 On Line Appendix The Golf Equipments Golf Bags Golf bags are designed to transport the golf clubs

More information

Setting up group models Part 1 NITP, 2011

Setting up group models Part 1 NITP, 2011 Setting up group models Part 1 NITP, 2011 What is coming up Crash course in setting up models 1-sample and 2-sample t-tests Paired t-tests ANOVA! Mean centering covariates Identifying rank deficient matrices

More information

Probabilistic Modelling in Multi-Competitor Games. Christopher J. Farmer

Probabilistic Modelling in Multi-Competitor Games. Christopher J. Farmer Probabilistic Modelling in Multi-Competitor Games Christopher J. Farmer Master of Science Division of Informatics University of Edinburgh 2003 Abstract Previous studies in to predictive modelling of sports

More information

Driv e accu racy. Green s in regul ation

Driv e accu racy. Green s in regul ation LEARNING ACTIVITIES FOR PART II COMPILED Statistical and Measurement Concepts We are providing a database from selected characteristics of golfers on the PGA Tour. Data are for 3 of the players, based

More information

Online Companion to Using Simulation to Help Manage the Pace of Play in Golf

Online Companion to Using Simulation to Help Manage the Pace of Play in Golf Online Companion to Using Simulation to Help Manage the Pace of Play in Golf MoonSoo Choi Industrial Engineering and Operations Research, Columbia University, New York, NY, USA {moonsoo.choi@columbia.edu}

More information

NBA TEAM SYNERGY RESEARCH REPORT 1

NBA TEAM SYNERGY RESEARCH REPORT 1 NBA TEAM SYNERGY RESEARCH REPORT 1 NBA Team Synergy and Style of Play Analysis Karrie Lopshire, Michael Avendano, Amy Lee Wang University of California Los Angeles June 3, 2016 NBA TEAM SYNERGY RESEARCH

More information

MEASURING CONTROL DELAY AT SIGNALIZED INTERSECTIONS: CASE STUDY FROM SOHAG, EGYPT

MEASURING CONTROL DELAY AT SIGNALIZED INTERSECTIONS: CASE STUDY FROM SOHAG, EGYPT MEASURING CONTROL DELAY AT SIGNALIZED INTERSECTIONS: CASE STUDY FROM SOHAG, EGYPT Ibrahim H. Hashim 1, Talaat A. Abdel-Wahed 2 and Ahmed M. Mandor 3 1 Associate Prof., Civil Eng. Dept., Faculty of Engineering,

More information

Title: Modeling Crossing Behavior of Drivers and Pedestrians at Uncontrolled Intersections and Mid-block Crossings

Title: Modeling Crossing Behavior of Drivers and Pedestrians at Uncontrolled Intersections and Mid-block Crossings Title: Modeling Crossing Behavior of Drivers and Pedestrians at Uncontrolled Intersections and Mid-block Crossings Objectives The goal of this study is to advance the state of the art in understanding

More information

Descriptive Statistics

Descriptive Statistics Descriptive Statistics Descriptive Statistics vs Inferential Statistics Describing a sample Making inferences to a larger population Data = Information but too much information. How do we summarize data?

More information

DEVELOPMENT OF A SET OF TRIP GENERATION MODELS FOR TRAVEL DEMAND ESTIMATION IN THE COLOMBO METROPOLITAN REGION

DEVELOPMENT OF A SET OF TRIP GENERATION MODELS FOR TRAVEL DEMAND ESTIMATION IN THE COLOMBO METROPOLITAN REGION DEVELOPMENT OF A SET OF TRIP GENERATION MODELS FOR TRAVEL DEMAND ESTIMATION IN THE COLOMBO METROPOLITAN REGION Ravindra Wijesundera and Amal S. Kumarage Dept. of Civil Engineering, University of Moratuwa

More information

PREDICTING the outcomes of sporting events

PREDICTING the outcomes of sporting events CS 229 FINAL PROJECT, AUTUMN 2014 1 Predicting National Basketball Association Winners Jasper Lin, Logan Short, and Vishnu Sundaresan Abstract We used National Basketball Associations box scores from 1991-1998

More information

Failure Data Analysis for Aircraft Maintenance Planning

Failure Data Analysis for Aircraft Maintenance Planning Failure Data Analysis for Aircraft Maintenance Planning M. Tozan, A. Z. Al-Garni, A. M. Al-Garni, and A. Jamal Aerospace Engineering Department King Fahd University of Petroleum and Minerals Abstract This

More information

Predicting the Total Number of Points Scored in NFL Games

Predicting the Total Number of Points Scored in NFL Games Predicting the Total Number of Points Scored in NFL Games Max Flores (mflores7@stanford.edu), Ajay Sohmshetty (ajay14@stanford.edu) CS 229 Fall 2014 1 Introduction Predicting the outcome of National Football

More information

THE INTEGRATION OF THE SEA BREAM AND SEA BASS MARKET: EVIDENCE FROM GREECE AND SPAIN

THE INTEGRATION OF THE SEA BREAM AND SEA BASS MARKET: EVIDENCE FROM GREECE AND SPAIN THE INTEGRATION OF THE SEA BREAM AND SEA BASS MARKET: EVIDENCE FROM GREECE AND SPAIN Lamprakis Avdelas, Managing Authority for the Operational programme for Fisheries, lamprakisa@in.gr Jordi Guillen, University

More information

COMPLETING THE RESULTS OF THE 2013 BOSTON MARATHON

COMPLETING THE RESULTS OF THE 2013 BOSTON MARATHON COMPLETING THE RESULTS OF THE 2013 BOSTON MARATHON Dorit Hammerling 1, Matthew Cefalu 2, Jessi Cisewski 3, Francesca Dominici 2, Giovanni Parmigiani 2,4, Charles Paulson 5, Richard Smith 1,6 1 Statistical

More information

Calculation of Trail Usage from Counter Data

Calculation of Trail Usage from Counter Data 1. Introduction 1 Calculation of Trail Usage from Counter Data 1/17/17 Stephen Martin, Ph.D. Automatic counters are used on trails to measure how many people are using the trail. A fundamental question

More information

Determining bicycle infrastructure preferences A case study of Dublin

Determining bicycle infrastructure preferences A case study of Dublin *Manuscript Click here to view linked References 1 Determining bicycle infrastructure preferences A case study of Dublin Brian Caulfield 1, Elaine Brick 2, Orla Thérèse McCarthy 1 1 Department of Civil,

More information

PGA Tour Scores as a Gaussian Random Variable

PGA Tour Scores as a Gaussian Random Variable PGA Tour Scores as a Gaussian Random Variable Robert D. Grober Departments of Applied Physics and Physics Yale University, New Haven, CT 06520 Abstract In this paper it is demonstrated that the scoring

More information

1. Answer this student s question: Is a random sample of 5% of the students at my school large enough, or should I use 10%?

1. Answer this student s question: Is a random sample of 5% of the students at my school large enough, or should I use 10%? Econ 57 Gary Smith Fall 2011 Final Examination (150 minutes) No calculators allowed. Just set up your answers, for example, P = 49/52. BE SURE TO EXPLAIN YOUR REASONING. If you want extra time, you can

More information

The probability of winning a high school football game.

The probability of winning a high school football game. Columbus State University CSU epress Faculty Bibliography 2008 The probability of winning a high school football game. Jennifer Brown Follow this and additional works at: http://csuepress.columbusstate.edu/bibliography_faculty

More information

The risk assessment of ships manoeuvring on the waterways based on generalised simulation data

The risk assessment of ships manoeuvring on the waterways based on generalised simulation data Safety and Security Engineering II 411 The risk assessment of ships manoeuvring on the waterways based on generalised simulation data L. Gucma Maritime University of Szczecin, Poland Abstract This paper

More information

EFFICIENCY OF TRIPLE LEFT-TURN LANES AT SIGNALIZED INTERSECTIONS

EFFICIENCY OF TRIPLE LEFT-TURN LANES AT SIGNALIZED INTERSECTIONS EFFICIENCY OF TRIPLE LEFT-TURN LANES AT SIGNALIZED INTERSECTIONS Khaled Shaaban, Ph.D., P.E., PTOE (a) (a) Assistant Professor, Department of Civil Engineering, Qatar University (a) kshaaban@qu.edu.qa

More information

5.1 Introduction. Learning Objectives

5.1 Introduction. Learning Objectives Learning Objectives 5.1 Introduction Statistical Process Control (SPC): SPC is a powerful collection of problem-solving tools useful in achieving process stability and improving capability through the

More information

A Comparison of Trend Scoring and IRT Linking in Mixed-Format Test Equating

A Comparison of Trend Scoring and IRT Linking in Mixed-Format Test Equating A Comparison of Trend Scoring and IRT Linking in Mixed-Format Test Equating Paper presented at the annual meeting of the National Council on Measurement in Education, San Francisco, CA Hua Wei Lihua Yao,

More information

Guidelines for Applying Multilevel Modeling to the NSCAW Data

Guidelines for Applying Multilevel Modeling to the NSCAW Data Guidelines for Applying Multilevel Modeling to the NSCAW Data by Sharon Christ, Paul Biemer and Christopher Wiesen Odum Institute for Research in Social Science May 2007 Summary. This document is intended

More information

STANDARDIZED CATCH RATES OF BLUEFIN TUNA, THUNNUS THYNNUS, FROM THE ROD AND REEL/HANDLINE FISHERY OFF THE NORTHEAST UNITED STATES DURING

STANDARDIZED CATCH RATES OF BLUEFIN TUNA, THUNNUS THYNNUS, FROM THE ROD AND REEL/HANDLINE FISHERY OFF THE NORTHEAST UNITED STATES DURING SCRS/2008/088 STANDARDIZED CATCH RATES OF BLUEFIN TUNA, THUNNUS THYNNUS, FROM THE ROD AND REEL/HANDLINE FISHERY OFF THE NORTHEAST UNITED STATES DURING 1980-2007 by Craig A. Brown 1 Sustainable Fisheries

More information

Comparisons of Discretionary Lane Changing Behavior

Comparisons of Discretionary Lane Changing Behavior THE UNIVERSITY OF TEXAS AT EL PASO Comparisons of Discretionary Lane Changing Behavior Matthew Vechione 1 E.I.T., Ruey Cheu, Ph.D., P.E., M.I.T.E. 1 Department of Civil Engineering, The University of Texas

More information

GLMM standardisation of the commercial abalone CPUE for Zones A-D over the period

GLMM standardisation of the commercial abalone CPUE for Zones A-D over the period GLMM standardisation of the commercial abalone for Zones A-D over the period 1980 2015 Anabela Brandão and Doug S. Butterworth Marine Resource Assessment & Management Group (MARAM) Department of Mathematics

More information

Discussion on the Selection of the Recommended Fish Passage Design Discharge

Discussion on the Selection of the Recommended Fish Passage Design Discharge Discussion on the Selection of the Recommended Fish Passage Design Discharge Introduction The provision of fish passage is a requirement for most proposed culvert and bridge installations in Alberta, depending

More information

Effects of Incentives: Evidence from Major League Baseball. Guy Stevens April 27, 2013

Effects of Incentives: Evidence from Major League Baseball. Guy Stevens April 27, 2013 Effects of Incentives: Evidence from Major League Baseball Guy Stevens April 27, 2013 1 Contents 1 Introduction 2 2 Data 3 3 Models and Results 4 3.1 Total Offense................................... 4

More information

Quantifying Bullwhip Effect and reducing its Impact. Roshan Shaikh and Mudasser Ali Khan * ABSTRACT

Quantifying Bullwhip Effect and reducing its Impact. Roshan Shaikh and Mudasser Ali Khan * ABSTRACT South Asian Journal of Management Sciences Vol. 1, No. 1, (Spring 2007) 25-31 Quantifying Bullwhip Effect and reducing its Impact Roshan Shaikh and Mudasser Ali Khan * ABSTRACT This paper quantifies factors

More information

A IMPROVED VOGEL S APPROXIMATIO METHOD FOR THE TRA SPORTATIO PROBLEM. Serdar Korukoğlu 1 and Serkan Ballı 2.

A IMPROVED VOGEL S APPROXIMATIO METHOD FOR THE TRA SPORTATIO PROBLEM. Serdar Korukoğlu 1 and Serkan Ballı 2. Mathematical and Computational Applications, Vol. 16, No. 2, pp. 370-381, 2011. Association for Scientific Research A IMPROVED VOGEL S APPROXIMATIO METHOD FOR THE TRA SPORTATIO PROBLEM Serdar Korukoğlu

More information

Forecasting sports tournaments by ratings of (prob)abilities

Forecasting sports tournaments by ratings of (prob)abilities Forecasting sports tournaments by ratings of (prob)abilities Achim Zeileis, Christoph Leitner, Kurt Hornik http://eeecon.uibk.ac.at/~zeileis/ UEFA Euro 2016 prediction Source: UEFA, Wikipedia UEFA Euro

More information

Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments of Finite Sample Properties

Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments of Finite Sample Properties Random Coefficient Models for Time-Series Cross-Section Data: Monte Carlo Experiments of Finite Sample Properties Nathaniel Beck Jonathan N. Katz Draft of April 1, 2006 We gratefully acknowledge the financial

More information

An Analysis of Factors Contributing to Wins in the National Hockey League

An Analysis of Factors Contributing to Wins in the National Hockey League International Journal of Sports Science 2014, 4(3): 84-90 DOI: 10.5923/j.sports.20140403.02 An Analysis of Factors Contributing to Wins in the National Hockey League Joe Roith, Rhonda Magel * Department

More information

Unit 4: Inference for numerical variables Lecture 3: ANOVA

Unit 4: Inference for numerical variables Lecture 3: ANOVA Unit 4: Inference for numerical variables Lecture 3: ANOVA Statistics 101 Thomas Leininger June 10, 2013 Announcements Announcements Proposals due tomorrow. Will be returned to you by Wednesday. You MUST

More information

Roundabout Design 101: Roundabout Capacity Issues

Roundabout Design 101: Roundabout Capacity Issues Design 101: Capacity Issues Part 2 March 7, 2012 Presentation Outline Part 2 Geometry and Capacity Choosing a Capacity Analysis Method Modeling differences Capacity Delay Limitations Variation / Uncertainty

More information

Returns to Skill in Professional Golf: A Quantile Regression Approach

Returns to Skill in Professional Golf: A Quantile Regression Approach International Journal of Sport Finance, 2010, 5, 167-180, 2010 West Virginia University Returns to Skill in Professional Golf: A Quantile Regression Approach Leo H. Kahane 1 1 Providence College Leo H.

More information